Carlos Saenz
University of South Florida Tampa, FL 33620
Carlos D. Paternina Arboleda
University of South Florida Tampa, FL 33620
Suresh K. Khator
University of South Florida Tampa, FL 33620
Resumo
This paper evaluates the performance of multiple-load automated guided vehicles (AGVs) under different conditions, and compares those results against the ones obtained with a single-load vehicle system. We design and conduct a full factorial analysis of variance on several response variables, namely flowtime, production completion, work-in-process, AGVs loaded utilization, and AGVs total utilization. We first run an experiment where load pickup and drop-off rules, vehicle fleet size, shop loading, and buffer capacity are the factors to be analyzed. In addition, we chose the best set of rules for the multiple-load vehicle system and compared their performance against the unit-load system. We found that the FSNS-a (First-Serve-Next-Station with priority to parts arriving to a queue) pickup rule performs best for a multiple-load vehicle, in terms of flowtime, WIP (Work In Process,) and AGVs total utilization. Only one performance measure was sensitive to drop-off rules, namely loaded AGVs utilization. The ND (Nearest Destination) drop-off rule minimizes this response variable. When we compared the multiple-load AGV system using the FSNS-a and ND rules against a unit-load system, we found out that the multiple-load system with only three vehicles performed better than the unit-load system with a fleet of four units. Larger buffer capacities were useful on the multiple-load system working at higher shop loading. Besides producing a contraction on flowtime, WIP inventory was also reduced. Results on the unit-load system did not improve when buffers are augmented.